Personalized medicine: An interview with Esther Dyson

Esther Dyson says we can expect to see a lot of changes in the personalization of health and health data.

Esther Dyson speaks about her participation in the Personal Genome Project, an initiative that aims to build and correlate genetic databases and personal risk factors, and 23andMe, her commercial venture that offers consumers the ability to read and understand their DNA. McKinsey’s director of publishing, Rik Kirkland, conducted this interview with Esther Dyson in April 2009.

Rik Kirkland: We are here today with Esther Dyson, who has a wonderful range of things she’s up to these days. A lot of companies you are investing in through EDventure, which is the firm you’ve created. You have put your genome up at the Personal Genome Project for the world to see, along with your health records. And you’re very excited, among other things, about the future of the potential of information technology and biotechnology to change the face of health care. So, thanks for being with us today. And I’d like to hear a bit about Esther’s excellent adventures in health care.

Esther Dyson: The biggest point, in a sense, is we need to focus not just on health care but on health. And that ranges all the way from helping people to change their own behavior—so that they end up healthier and need less care and also enjoy more health—to going to the bowels of the health care system and helping hospitals to become more efficient and avoid mistakes.

In another direction, there’s the kinds of things that more genetic information will enable us to do in creating better and more targeted drugs. Ryan Phelan, [founder and CEO of] DNA Direct, said to me once, “You’d no more think of getting a drug without knowing your genome than you’d think of getting a blood transfusion without knowing your blood type.”

Rik Kirkland: Shifting from delivering volumes of outputs, care, to actually creating health at the personal level—sort of the common thread of both these things?

Esther Dyson: Right. And everything the Internet tells you about personalization and personal data in comparing yourself to other people. And mass marketing turning into the market of one. That’s what we’re going to see in both health and health care.

Rik Kirkland: Talk a little bit about the genomics side and tell us why you did this and what you hope to get out of that. And of course, you’re also involved in the board and as an investor in 23andMe. So, tell us a little bit about that.

Esther Dyson: The idea behind the Personal Genome Project was created by George Church at Harvard Medical School. It begins with ten people who volunteer to put their entire genome and all their health records online for anybody to see, with our identities visible. And the idea there was, in a sense, to demystify it. And then, ultimately, to [map] hundreds of thousands of people, so that we’d have a lot more data to [use to] explore the genome and medical care and so forth and so on. Most medical research, certainly when it’s published, is with unidentified subjects. And the idea here was for us to be role models, to prove that this information wasn’t secret, or scary, or dangerous.

Rik Kirkland: Part of the barrier to getting electronic records is all the concern about releasing this stuff and all the forms you sign every time you go to the doctor. So, this is just a radical act of transparency, right?

Esther Dyson: Right. And you know, you go to any health insurance company, they don’t want all that detail. They certainly want to know about preexisting conditions, but you don’t need to know your genome to know those. And the genome itself was too much information and it doesn’t really tell you that much.

But if your mother died of a heart attack, or your father has colon polyps, as mine does, it’s pretty clear what the indications are. Genomes themselves give you only—with a few exceptions—percentages. So, it’s not like you put this information up and people are going to stick pins into it. I actually was expecting more medical spam about, you know, “We looked at your genome. You should buy such and such nutri-ceutical.” What will be exciting is when you have hundreds of thousands of these and you say, “Oh, wow. They’re these five genes that seem to interact.” Most things are not a gene. It’s usually a lot of different genes—and then, combined with what you eat and whether you sleep enough and whether you stay warm enough and all these other things—that actually produce a real outcome in a person of being in such and such condition.

Rik Kirkland: So, how far are we away from this kind of personalized medicine that knowing your genome and knowing more about the patterns the genome reveals. How far away is that moment?

Esther Dyson: Well, it’s like everything else. A few things we already know. Different types of breast cancer have different prognoses, different recommended treatments, that kind of thing. I think we’re pretty close to getting there on, sort of, overall things, like how fast you metabolize drugs in general and specific drugs. Originally, type 1 and type 2 diabetes, that was not a genetic discovery, but it was the same kind of discovery. “Oh, it looks like the same condition, but really they are two different causes and very different patterns.” And so forth.

We’re there in places, but we’re very, very far from being there in general. This is something that people should want to do or volunteer to do, not something that should be imposed on them.

Rik Kirkland: You said earlier that the harder thing was not getting your genome done—although that’s expensive right now and will get less so as the process scales—but just getting your medical records together. So, talk a little bit about the information technology possibilities and finally get our arms around this medical records piece.

Esther Dyson: There are different things. One, individuals assembling their own health record. Suddenly, when any doctor treats you, he can know what the other doctors have done, what drugs you were taking, what conditions you might have—your genetic information could be helpful, but it’s just one piece. And so, no doctor, especially in these days of specialists, can know everything that’s done to you. Nor can every doctor know all the medical knowledge that might apply to you and your condition. Everyone else uses electronic memory aides, why shouldn’t doctors?

There’s this sort of macho/priesthood notion in doctors that they should know everything. But there are companies now that are already using drug-interaction databases. And those are very useful in averting tragedies all the time, where patients take 2 or 3 drugs from different doctors and it kills them. More often, you just have somebody who’s taking 18 or 19 different drugs. And you take 10 of them away, and the patient gets better rather than worse. So, that’s the first part, just the patient having the data aggregated in one place.

Then you have, of course, the longitudinal thing—watching how things change over time, seeing the patient’s reaction to drugs. And if you aggregate all that information across lots of patients, lots of genomes, lots of treatments, suddenly you begin to see both correlations. And then you can actually figure out how the different forces interact. And that’s when it gets really exciting.

In the meantime, one hopes also that hospitals—and all kinds of care providers—will get more efficient. They won’t have to redo tests. They’ll make better treatment decisions. You’ll be able to reward people for better outcomes rather than simply pay for health care. What you really want to pay for is health, which is the difference between the expected outcome and a better outcome.

Rik Kirkland: As you said, the ultimate game here is better health, personalized health, from all this amassing of data, and then being able to apply it to your particular situation.

Esther Dyson: But there’s another impact, that if you know more about yourself and how things work, on the margin, chances are, you’re going to be more healthy. You’ll be better armed against the doughnut. Or maybe not the extra drink, that’s very seductive.

Rik Kirkland: Don’t underestimate doughnuts. They’re seductive too.

Esther Dyson: The stuff people eat is crazy. And they know it, sort of. It’s important if they know it scientifically, but also if they know it in relation to themselves. Businesses do projections all the time. You continue this spending rate, you’ll run out of cash next February. You continue this behavior rate, you’re probably going to get a heart attack before you’re 60.

That kind of specific prediction is really good in changing behavior. And so are reminders, so are social networks where patients, where healthy people encourage one another to exercise or remember to take their meds or just find comfort from people who are suffering from the same problem you have.

Rik Kirkland: Ten years from now, will we see a really dramatic transformation, do you think, in how health care is delivered? Will it really be far more personalized? Is it a 20-year process?

Esther Dyson: I think you’ll see a lot of changes in five to ten years. But not for everybody. Like everything else, this is going to benefit—let’s face it—the rich and the well educated first. They’re the ones right now who, if they have a problem, they probably have a friend at Mount Sinai [Hospital]. And so, they can get the best care anyway. You shouldn’t need a friend at Mount Sinai to get good cancer care. That’s the real difference. There will still be a leg for people who live in the wrong place or who are at the bottom of the pyramid. When you do things with computers and with IT, they actually scale very well. So, you can’t reproduce the friend at Mount Sinai. But you can reproduce an information system that gives you access to the same information.

Rik Kirkland: That was formerly only there for the friend in Mount Sinai, and now you can get it cheaper.

Esther Dyson: Yes, right.

Rik Kirkland: So, that’s the hope over a 10- to 20-year period?

Esther Dyson: Yes, that’s the hope.

Rik Kirkland: Great. Thank you, Esther.